Article Text

Download PDFPDF

Original research
Assessment of intra-aneurysmal flow modification after flow diverter stent placement with four-dimensional flow MRI: a feasibility study
  1. Vitor Mendes Pereira1,2,3,
  2. Olivier Brina1,
  3. Benedicte M A Delattre4,
  4. Rafik Ouared1,
  5. Pierre Bouillot1,
  6. Gorislav Erceg1,
  7. Karl Schaller5,
  8. Karl-Olof Lovblad1,
  9. Maria-Isabel Vargas1
  1. 1Division of Neuroradiology, Department of Medical Imaging, University Hospitals of Geneva, Geneva, Switzerland
  2. 2Division of Neuroradiology, Department of Medical Imaging, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada
  3. 3Division of Neurosurgery, Department of Surgery, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada
  4. 4Division of Radiology, Department of Medical Imaging, University Hospitals of Geneva, Geneva, Switzerland
  5. 5Division of Neurosurgery, University Hospitals of Geneva, Geneva, Switzerland
  1. Correspondence to Dr Vitor Mendes Pereira, Division of Neuroradiology, Department of Medical Imaging and Division of Neurosurgery, Department of Surgery, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada; vitormpbr{at}hotmail.com

Abstract

Background Flow diverter stents (FDS) have been effectively used for the endovascular treatment of sidewall intracranial aneurysms (IAs). Unlike standard endovascular treatments used to exclude directly the aneurysm bulge from the parent vessel, FDS induce reduction in the intra-aneurysmal flow and promote progressive and stable thrombosis therein. The advent of FDS has therefore increased the need for understanding of IA hemodynamics.

Methods We proposed the use of the most recently evolved four-dimensional (4D) flow MRI technique to evaluate qualitatively and quantitatively post-FDS flow modification in 10 patients. We report intra-aneurysmal velocity measurements and the influence of metal artifacts induced by the stent.

Results An index was defined to quantitatively measure flow changes—namely, the proportional velocity reduction ratio (PVRR)—with ranges from 34.6% to 71.1%. Furthermore, we could compare streamlines characterizing the post-stent flow patterns in five patients in whom the intra-aneurysmal velocity was beyond the visualization threshold of 7.69 cm/s.

Conclusions Despite metal artifacts and the low velocities involved, 4D flow MRI could be of interest to measure qualitatively and quantitatively flow changes in stented aneurysms. However, further enhancements are required together with further validation work before it can be considered for clinical use.

  • Aneurysm
  • Blood Flow
  • Flow Diverter
  • MRI
  • Stent

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Intracranial aneurysms (IAs) can evolve as life-threatening intracranial hemorrhages.1–3 Large aneurysms present a greater risk of rupture and usually require preventive treatment.1 ,4 Among different endovascular techniques such as coiling with or without stent assistance, flow diverter stents (FDS) have been successfully used to treat large neck and complex aneurysms.5 Unfortunately, FDS have also been related to some post-procedural complications such as subacute aneurysm rupture, parent artery occlusion, and persistence of aneurysm flow despite treatment.6 Although many hypotheses have been proposed to explain some of these complications, most of the mechanisms of action of FDS remain unknown. To date there is no method for measuring flow modification reliably and predicting safe and effective aneurysm occlusion after FDS. Digital subtraction angiography (DSA) using an optical flow-based algorithm has recently been used to evaluate peroperative intra-aneurysm flow changes after FDS placement and hence the correlation with aneurysm thrombosis.7 ,8 The initial results showed that this method significantly predicted thrombosis issues correctly in 24 patients. However, DSA requires an invasive approach which precludes its extensive use during follow-up of patients with FDS placement. Blood flow simulation studies using computational fluid dynamics (CFD) have been used to evaluate flow modification after virtual deployment of FDS9 but have not achieved clinical relevancy. In particular, the porosity of the device in relation to vessel curvature together with geometrical variations in the parent vessel after deployment are technical issues that are still not taken into account with CFD models.

Four-dimensional (4D) phase contrast MRI (4D PC-MRI) has evolved as a method to assess qualitatively and to measure quantitatively the hemodynamics for cardiac and aortic applications.10 ,11 Higher magnetic fields and flow sequence improvements enable 4D flow MRI to access vascular structures with smaller diameters such as intracranial vessels.12–15 This technique has been used to evaluate the hemodynamics of IAs, and preliminary validation work using in vitro experiments and computational models has been performed.16–22 It is well known that endovascular implants are related to metal artifacts and signal void issues depending on the metallic composition of the alloy.23 However, Bunks et al experimentally showed that 4D flow imaging could be used to measure intra-stent flow in different types of peripheral stents.24 Karacozoff et al25 reported that surpass NeuroEndograft FDS-related artifacts may be problematic if the region of interest is in the same area or close to the implant. In the same study they also reported that MRI of the same device presents no concerns with regard to heating and magnetic field interactions.

We therefore suggest that application of 4D flow sequences could be used in MRI follow-up of patients treated with the same kind of device. In this study we evaluate the feasibility of 4D flow MRI to assess intra-aneurysmal flow modifications after FDS implantation as well as its potential drawbacks and limitations.

Materials and methods

Patients and FDS implantation

Ten consecutive patients with unruptured IAs treated with FDS were included in the study. The aneurysms were all located in the internal carotid artery (ICA) in the cavernous portion (n=4), posterior communicating artery segment (n=1), or ophthalmic portion (n=5). We used three different FDS: Pipeline embolization device (PED; Covidien Neurovascular, Irvine, California, USA), Silk stent (Balt, Montgomery, France) and FRED device (Microvention Terumo, Irvine, California, USA). The choice of FDS was based on navigability and implantation issues. Our treatment strategy was to implant a single FDS layer, except in cases in which a telescoping technique was needed to extend the effective length of stent coverage. No combination of coil was considered in the first attempt. The patients were treated under general anesthesia. Using the Seldinger approach, 8 F femoral access was achieved for all patients. An 8 F guiding catheter was then used as a support system coaxially with a 6 F Navien catheter (Covidien Neurovascular) or a Fargo Max catheter (Balt). The specific microcatheter was navigated distal to the aneurysm neck and the FDS was deployed according to the individual anatomy and device delivering technique. Peroperative VasoCT (XperCT, Philips Health Care, Best, The Netherlands) was performed after deployment to ensure optimal device apposition to the vessel wall.

MRI

We performed two flow MRI examinations, the first on the day before the procedure and the second during the 48 h following the procedure (as soon as the patient's condition ensured a safe and good quality MR examination). The post-procedure MRI was performed as soon as possible after implantation to avoid potential onset of thrombosis and to ensure that hemodynamic changes were due to the stent itself and not to regression of the aneurysm lumen. MRI was performed using a Philips Ingenuity TF PET/MR with an Achieva 3.0T TX series MR system (Philips Healthcare, Cleveland, Ohio, USA). The following acquisition protocol was used for both examinations. First, a 3D time of flight (TOF) sequence was acquired in transverse orientation with the following parameters: FOV 200×200 mm2, 200 slices, acquisition voxel size 0.42×0.73×0.55, reconstructed with 0.3 mm3 isotropic resolution, SENSE acceleration factor 2.5, TR/TE 25/3.5 ms. The 4D PC-MRI sequence was then placed and orientated based on the transverse 3D TOF slices and their MIP reconstructions in axial, sagittal and coronal orientations, enclosing the entire aneurysm volume and the adjacent parent vessel. The flow sequence encompassed velocity encoding in the three directions with the following parameters: FOV 190×210 mm2, typically 32 slices (adjusted according to aneurysm size), acquisition voxel size 1×1×1 mm3 reconstructed with 0.82 mm3, sense acceleration factor 2, TR/TE 4.6/2.9 ms. For a typical heart rate of 65 bpm, the number of cardiac phases was 16 and the acquisition time was approximately 13 min. Cardiac triggering was performed with a peripheral pulse unit. The velocity sensitivity encoding parameter (VENC) was set to 80 cm/s to avoid as far as possible aliasing artifacts without degrading the signal-to-noise ratio, as recommended for intracranial arteries by Markl et al.12 An eddy current correction was performed during the reconstruction of the phase images. Regular clinical sequences were also performed for each pre- and post-examination (T2-weighted turbo spin echo, T2-weighted gradient echo and diffusion weighted imaging).

Image analysis

Dynamic flow sequences were visualized and qualitatively evaluated using GT-Flow software (GyroTools, Zurich, Switzerland). Based on the 4D PC-MRI flow data, streamlines were calculated within a threshold-segmented volume and from additional regions of interest (ROIs) within the aneurysm. These streamlines were used for qualitative assessment over the 16 time steps covering the cardiac cycle. Intra-aneurysmal flow features underlined by streamlines were described as visible or not for the pre- and post-procedure flow data. If post-procedure streamlines were visible, a comparison with the pre-implantation data was performed. Furthermore, the presence of intra-stent streamlines was documented.

Quantitative assessment

We defined, on pre- and post-procedure flow data, aneurysm ROIs to perform a quantitative evaluation of the velocity reduction after FDS placement. Because the patients’ head positions were slightly different between the pre- and post-procedure MRI, the ROIs were manually delineated on equivalent slices visualized in velocity magnitude and were restricted to the internal contour of the aneurysm, with special attention paid to avoiding contamination by the parent vessel velocity components and far enough from stent artifacts, as recommended by Karacozoff et al.25 The positioning of the ROI on the aneurysm is shown in figure 1. Additionally, we defined ROIs in the C4 segment of the ICA for both flow datasets. The maximum value of velocity within the 16 cardiac phases was extracted for both the aneurysm (Velan) and the ICA (Velart). We then defined pre- and post-stent (S0 and S1 respectively) maximum velocity ratios (Ran and Rart) to account for possible variations in ICA velocities between the two examinations. Finally, a proportional velocity reduction ratio (PVRR) was calculated, reflecting the velocity reduction after stenting normalized to the parent vessel velocities (equation 1).Embedded Image 1

Figure 1

Regions of interest (ROI) for patient 9 who presented with a left carotid ophthalmic aneurysm. A and B are magnitude images with and without ROI. C, D and E are phase contrast images in x, y and z velocity encoded directions, respectively. Top and bottom rows show pre- and post-implantation data, respectively. In the bottom row the signal void artifact due to the presence of the stent can be seen, which still allows visualization of aneurysm flow.

We also measured the linear correlation (R2) of PVRR with the following morphological parameters: aneurysm volume, size ratio (maximum height over inlet vessel size), aspect ratio (height to neck ratio), parent vessel diameter, aneurysm size (height), and nominal stent length coverage over neck length.

Results

FDS were successfully implanted in the 10 patients without perioperative complications. Table 1 summarizes the geometrical features of the 10 aneurysms. The aneurysm volumes ranged from 0.054 to 1.817 cm3. For nine patients the post-implantation MRI was performed within 2 days following the implantation and no signs of early thrombosis were observed with regular sequences. One patient (patient 8) presented with a mental status disorder after the procedure and could not undergo MRI until 9 days after implantation. This patient showed signal characteristics compatible with an already partially thrombosed aneurysm on post-procedure regular sequences, precluding any evaluation of intra-aneurysmal flow. Additionally, a telescopic technique was used for patient 5, who was implanted with two PEDs.

Table 1

Patient and aneurysm geometrical parameters. These parameters were measured on 3D reconstructions acquired during intraprocedural angiography

Stent-induced artifacts

Magnetic susceptibility artifacts were observed on magnitude and phase contrast images for all of the patients and all of the FDS. These artifacts were primarily located at the ends of the stents and, to a lesser extent, close to their extraluminal and intraluminal surfaces. The level of these artifacts was different among the patients and was influenced by device orientation, related to the longitudinal axis of the magnetic field (B0), rather than by the device itself. This feature has already been described by Klemm et al26 in in vitro experiments, and we also found that these susceptibility artifacts were attenuated when the orientation of the stent was probably parallel to B0. In five patients an intra-stent signal was present despite the influence of stent artifacts; the other five patients had no signal. We observed that the presence of artifacts was more significant with the telescopic PED in case 5 (double strut layers) and with the FRED stent, which is built as a double layer device.

Aneurysm streamline visualization

The pre-stent flow datasets presented overall good imaging quality and provided intra-aneurysmal flow for all 10 patients. It was indeed possible to describe quite unambiguously the direction of the inflow stream, as well as the intra-aneurysmal recirculation patterns. Concerning the post-implantation data, we were able to analyze intra-aneurysmal flow structures for five of the 10 patients (table 2 summarizes the results for each patient), even if streamline pathways were difficult to follow and consequently misleading at the aneurysm dome where the velocities are expected to be low. Among these cases, two presented with significantly different flow behaviors between the pre- and post-implantation data. For instance, patient 3 (who had an ICA aneurysm in the ophthalmic segment) showed an interesting flow pattern modification induced by the implantation of a 4.75/16 mm PED (figure 2). Preoperatively, it could be observed that the flow of the ICA went through the aneurysm to create a clockwise vortex before entering the distal part of the ICA. After FDS placement the flow originating from the ICA was redirected within the stent, leaving the aneurysmal inflow stream reduced in size and velocity. We observed the same interesting patterns on DSA images, which illustrated the behavior of the flow at different time points. For the remaining three cases in which post-procedure streamlines were analyzable, the flow behavior was not significantly modified by the FDS. The direction of the inflow stream and intra-aneurysmal recirculation was identical. However, the velocity magnitude of the inflow stream and the aneurysmal recirculation were reduced compared with the preoperative data. Patient 6, who presented with a right ICA cavernous aneurysm which was implanted with a 4.75/25 mm PED, showed a very similar behavior after FDS implantation but with a lower velocity magnitude (figure 3). Finally, in four patients no intra-aneurysmal flow patterns were visualized after FDS implantation.

Table 2

List of implanted devices

Figure 2

Patient 3 presented with a right carotid ophthalmic aneurysm which was treated with a Pipeline embolization device 4.75/16. Top row: 3D reconstruction in AP, superior and PA projections. Rows 2 and 3: streamline visualization and four different time steps of digital subtraction angiography in AP projection before implantation of flow diverter stent. In these two rows the overall flow coming from the right carotid artery (RICA) goes through the aneurysm before entering the distal part of the RICA. Rows 4 and 5 are post-implantation images in which the inflow stream of the RICA is redirected within the stent, leaving an aneurysmal inflow with lower velocity. Intra-stent flow is clearly visible, even when only slightly altered.

Figure 3

Patient 6 presented with a right internal carotid artery cavernous aneurysm which was treated with a Pipeline embolization device 4.75/25. Top row: 3D reconstruction in lateral and AP projections. Rows 2 and 3: streamline visualization before and after stenting, respectively, in the same projection views. This case showed the same shape of flow patterns between the two datasets, with a lower magnitude of aneurysm velocity after stenting. Intra-stent flow is not visible.

Intra-stent streamlines

No arterial flow information could be observed in the five patients for whom the magnitude images showed significant changes in the intra-stent signal. However, it was possible to obtain intra-stent streamlines in five cases, three of which were altered at the diastolic phase of the cardiac cycle. It was noted that the absence of a signal and consequently of flow information seemed not to be related to FDS type. However, if the intra-stent signal was abolished, intra-aneurysmal streamlines were not necessarily altered (figure 3). It was also noted that, upstream and downstream of the stent, the streamlines were not disturbed.

Velocity reduction

Velocities were collected as aneurysm ROI-averaged values. Their maximum values within the 16 cardiac phases are summarized in table 2. In our patients, the aneurysms presented in a broad range of sizes resulting in pre-stent velocities varying widely from 13.65 to 45.27 cm/s. However, post-stent velocities were significantly reduced after stenting in all cases (p=0.005), although ICA velocities were not significantly different (p=0.114). After stenting, the four patients without visible intra-aneurysmal streamlines presented with maximum velocity magnitudes of less than 7.69 cm/s. This velocity level was the threshold beyond which intra-aneurysmal streamlines were assessable. PVRR, which expresses the velocity reduction after stenting, ranged from 34.6% to 71.1% for nine patients (excluding patient 8 who underwent delayed post-stent MR examination) and could be associated with a decrease in intra-aneurysmal flow after FDS. There was no significant linear correlation between PVRR and any of the morphological parameters tested (R2 <0.25). Moreover, PVRR was not related to FDS type. Overall, no velocity-aliasing artifacts were observed with the 80cm/s VENC.

Discussion

As an alternative to DSA, MRI is used for the follow-up of patients implanted with FDS. In addition to assessment of aneurysmal flow, MRI is also able to evaluate thrombosis and aneurysm resumption. However, susceptibility artifacts in the imaging of the stented vasculature remain an issue in evaluating the postoperative status, particularly with high material density prostheses.26 ,27 In particular, it was shown by Karacozoff et al25 that the imaging of the surpass FDS was safe but was altered by artifacts. However, Bunks et al showed that 4D flow imaging could be used to measure intra-stent flow in an in vitro experiment using different types of peripheral stents.24 They also showed that quantitative and qualitative assessment of the in-stent flow was reliable and could be used even in pathological conditions.

Our pilot study investigated the feasibility of MRI to evaluate intra-aneurysmal flow reductions after stenting using an acquisition protocol based on previous work conducted on cerebral aneurysm imaging.12 We considered two main technical issues associated with flow imaging after FDS placement: (1) the magnetic susceptibility of the material and associated signal void artifacts; and (2) the fact that the precise assessment of intra-aneurysmal flow reduction may not be compatible with the VENC settings. Despite these issues, we were able to perform a qualitative evaluation of intra-aneurysmal flow modification in five out of 10 patients using streamline visualization. The absence of post-stent flow features was due to very low intra-aneurysmal velocities induced by the stent (<7.69 cm/s), but this was not related to either the type of device or the aneurysm size. Furthermore, as shown in figure 1, interesting changes in flow features were found in which redirection of the upstream flow within the stent was observed.

Quantitatively, despite the range of aneurysm sizes, post-implantation velocity measurements could be obtained for nine patients including the four cases with weak streamline calculations. The measured PVRR ranged from 34.6% to 71.1% of velocity reduction, consistent with the computational studies conducted by Shobayashi et al and Kulcsar et al.28 ,29 These results were corroborated by a broad range of pre- and post-implantation velocities values, highlighting a number of factors linked to the patient and/or physician involved in the flow reduction effect: (1) anatomical aspects (lesion size, aspect ratio, aneurysm implantation on parent vessel); and (2) procedural factors (stent type, compression of the device during deployment, device oversizing, angioplasty). Morphological parameters had no significant impact on PVRR evaluation (R2 <0.25). Thus, PVRR was not affected by morphological biases and therefore solely reflected velocity reduction.

These flow MRI-based measurements must be confirmed and further validated in prospective studies with larger cohorts of patients in order to correlate these qualitative and quantitative features with relevant clinical issues such as post-procedure subacute rupture and time to thrombosis.

To date, two major approaches have been used to evaluate the flow modification effects induced by FDS. First, per-procedure indirect measurements using DSA were used by Pereira et al to develop an intra-procedure method using optical flow imaging to measure the flow reduction effects of FDS, which could significantly predict thrombosis outcomes.7 ,8 Struffert et al calculated quantifiable parameters based on time density curves to predict aneurysm occlusion in an animal study30 and Chien et al used a novel approach to map differences in blood flow motion between pre- and post-DSA sequences in four patients.31 However, even if these DSA-based methods of flow modification measurements could potentially support intraprocedural decisions, they remain difficult to apply during angiographic follow-up since the projection views need to be equivalent throughout the long-term evaluation, which is difficult to achieve between successive angiograms. Second, computational methods using CFD simulations have been applied.9 ,28 However, CFD remains a complex process which requires numerous assumptions, including challenging patient-specific modeling of the device. This approach is therefore a long way from future clinical application.

Even though a number of experimental and clinical validation steps are required before considering any clinical implications, we show that 4D flow MRI may have the potential to provide direct in vivo measurements of FDS-induced flow changes. The potential clinical insights provided by acquiring additional 4D flow MRI sequences during patient follow-up include assessment of qualitative modifications induced by the FDS and evaluation of the effective velocity reduction by quantitative measurement of pre-/post-stent indicators such as PVRR. Furthermore, 4D flow MRI could be used to monitor the flow reduction associated with thrombosis progression. In addition to these clinical insights, 4D flow MRI in the context of FDS could be of great interest for validating or cross-validating CFD studies. However, improvements in acquisition should be considered to obtain flow information for all configurations of aneurysms and stents. Essentially, sequence acceleration would provide access to smaller aneurysm characterization by increasing the spatial resolution with reasonable scanning times. Furthermore, simultaneous encoding with two different VENCs could offer improvements in the capture of slow and fast neighboring flows (‘dual VENC’ acquisitions).12

Some limitations of our study should be considered. The ROI used for velocity measurements was manually but carefully delineated in the pre- and post-stent datasets on equivalent slices, and it could have been positioned with slight differences. Knowing that IA velocities span a broad range, the location of the ROI could have influenced the measurement. However, manual delimitation is probably the safest option for determining intra-aneurysmal velocities without parent vessel contamination because our aneurysms closely neighbored the sinuous part of the carotid syphon. We identified a velocity threshold below which the streamlines were not visible (7.69 cm/s). This suggests that the measurement error is of this level of magnitude and that lower velocities cannot be reliably quantified. Consequently, the values of PVRR must be interpreted carefully in those cases where measured velocities were below this threshold. Our data involved a broad range of aneurysm sizes and topologies with their own hemodynamic classes as well as three different FDS types.

Finally, in vitro experiments should be performed to evaluate precisely the effect of stent-induced susceptibility artifacts in and around the location of the stent on the velocity measurements. No correlation was made between PVRR and treatment outcomes (time to thrombosis) since our study is a preliminary work in applying 4D flow MRI to patients with FDS.

Conclusion

In this feasibility study we have shown that 4D flow MRI can provide information on flow modification after FDS placement which is independent of stent type, despite problems of susceptibility artifacts and low velocities. Further developments are expected to improve flow sequences in evaluating intra-aneurysmal flow both qualitatively and quantitatively. Further experimental and clinical validation is needed before any clinical application of this method in patient follow-up can be considered.

References

Footnotes

  • Contributors Conception: VMP, OB, BMAD, M-IV. Clinical data: KS, K-OL, GE, VMP. Imaging review and follow-up: K-OL, OB, PB, RO. Post-processing: OB, PB, RO, BMAD. Writing: VMP, OB, BMAD, RO, PB. Review: VMP, OB, BMAD, RO, PB, GE, KS, K-OL, M-IV.

  • Funding Swiss National Funds (SNF 32003B_141192) and Swiss National Funds (SNF 320030_156813).

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Ethical committee and institutional board (NAC 12-064).

  • Provenance and peer review Not commissioned; externally peer reviewed.